Log inSign up
user avatar
Ameet Talwalkar
@atalwalkar
AI @ Datadog and CMU
Pittsburgh, PA
cs.cmu.edu/~atalwalk/
Joined October 2010
198
Following
2,598
Followers
  • Pinned
    user avatar
    Ameet Talwalkar
    @atalwalkar
    May 14
    Today weโ€™re releasing Toto 2.0: a family of open-weights time series foundation models spanning 4M to 2.5B parameters. The question we set out to answer was simple (yet previously open): Do time series foundation models get reliably better as they scale? Our answer: yes! ๐Ÿงต
    Image
    62K
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Jan 24, 2025
    I have some news to share! @datadoghq is forming a new AI research lab, and I'm excited to announce that I've joined as Chief Scientist to lead this effort. Datadog has a great work culture, lots of data and compute, and is committed to open science and open sourcing. Our team
    Image
    31K
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Feb 14, 2023
    Can we use LLMs to tackle genomics tasks? Vision transformers to solve PDEs? In new work led by @JunhongShen1, we consider the general problem of cross-modal fine-tuning and provide surprisingly optimistic answers to these questions. 1/N
    arXiv logo
    arxiv.org
    Cross-Modal Fine-Tuning: Align then Refine
    Fine-tuning large-scale pretrained models has led to tremendous progress in well-studied modalities such as vision and NLP. However, similar gains have not been observed in many other modalities...
    30K
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Jan 22, 2020
    Check out the exciting list of accepted papers (mlsys.org/Conferences/20โ€ฆ) for #MLSys20, March 2nd-4th in Austin, TX! Early registration ends on Feb 4th (mlsys.org)
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Nov 22, 2019
    SysML --> MLSys: Earlier this year SysML opted to change the conference name in response to trademark concerns. After receiving input from the Steering and Program Committees, we converged on MLSys. Looking forward to MLSys in March in Austin! #mlsys mlsys.org
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Mar 18, 2021
    What started as a series of brainstorming sessions to keep my group connected during the pandemic has resulted in a new survey paper on interpretable ML that highlights the need to ground the field in real problems and to quantifiably measure progress: arxiv.org/abs/2103.06254
    Image
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Jan 14, 2019
    SysML19 accepted papers just posted online (sysml.cc/papers.html). Registration opens on 1/16 at 10am PST. #sysml19
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Aug 18, 2022
    Image classifiers often rely on spurious patterns (eg, using presence of a person to detect a tennis racket) that can lead to problems with fairness, safety & distributional robustness. We introduce a method for finding & fixing such patterns that relies on model explanations.
    Image
    user avatar
    Accepted papers at TMLR
    @TmlrPub
    Aug 17, 2022
    Finding and Fixing Spurious Patterns with Explanations Gregory Plumb, Marco Tulio Ribeiro, Ameet Talwalkar openreview.net/forum?id=whJPuโ€ฆ
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Aug 30, 2021
    The call for papers for MLSys22 is now live (mlsys.org/Conferences/20โ€ฆ). Paper deadline is October 8th.
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Feb 20, 2024
    Congrats @gingsmith , @justinesherry , Aaditya Ramdas and Nathan Beckmann!! cs.cmu.edu/news/2024/sloaโ€ฆ
    5.6K
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Feb 21, 2019
    It turns out that random search works really well for neural architecture search...
    arXiv logo
    arxiv.org
    Random Search and Reproducibility for Neural Architecture Search
    Neural architecture search (NAS) is a promising research direction that has the potential to replace expert-designed networks with learned, task-specific architectures. In this work, in order to...
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Dec 12, 2018
    We just started a new CMU blog focused on machine learning! blog.ml.cmu.edu/2018/12/12/intโ€ฆ Really excited for students to showcase their amazing research, starting with my student Liam Li's post on massively parallel hyperparameter optimization blog.ml.cmu.edu/2018/12/12/masโ€ฆ #automl
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Apr 17, 2020
    Excited to share our new work on neural architecture search, joint with Liam, Misha (@khodakmoments), and Nina. New perspectives on designing weight-sharing methods, some simple theory, and state-of-the-art results on CIFAR/ImageNet NAS benchmarks
    arXiv logo
    arxiv.org
    Geometry-Aware Gradient Algorithms for Neural Architecture Search
    Recent state-of-the-art methods for neural architecture search (NAS) exploit gradient-based optimization by relaxing the problem into continuous optimization over architectures and shared-weights,...
  • user avatar
    Ameet Talwalkar
    @atalwalkar
    Sep 30, 2020
    1/ I recently began a podcast collaboration with @craigss to chat with friends and colleagues about various aspects of ML pipelines. Excited to release our first podcast focused on ML hardware, with Dave Patterson. You can listen to our conversation here: determined.ai/blog/dave-pattโ€ฆ